Increased Mortality Rates Associated with Staphylococcus aureus and Influenza Co-infection, Maryland and Iowa, USA

We retrospectively analyzed data for 195 respiratory infection patients who had positive Staphyloccocus aureus cultures and who were hospitalized in 2 hospitals in Iowa and Maryland, USA, during 2003–2009. Odds for death for patients who also had influenza-positive test results were >4 times higher than for those who had negative influenza test results.

like illness (ILI) (5). This criterion was part of an initial study investigating influenza-like illness and S. aureus pneumonia (J.S. McDanel, unpub. data). Patients were included in the study if they had respiratory cultures (sputum, bronchial specimen, or tracheal aspirate) that grew S. aureus and were tested for influenza before or during their admissions. If a patient was admitted >1 time, only the admission with the first S. aureus positive respiratory culture was included. The University of Iowa institutional review board approved this study.
The primary outcome of interest, 30-day in-hospital mortality, was defined as death occurring in the hospital within 30 days of the first culture that grew S. aureus. The adapted Charlson Comorbidity Index served as an aggregate score for co-occurring conditions (6). The year of each patient's first positive S. aureus culture was dichotomized: 2003-2007 and 2008-2009. We conducted bivariable analyses using either the χ 2 test or the Fisher exact test for categorical variables and the Student t-test or Wilcoxon rank-sum test for continuous variables. We used logistic regression to identify associations between potential predictor variables and 30-day mortality rates. We included variables in the multivariable model using a manual stepwise method. Variables associated with death (p<0.25) in the bivariable regression analysis were examined for fit within the multivariable model and were retained if statistically significant (p<0.05). The year of each patient's first positive S. aureus culture was forced into the model. We analyzed data using SAS software version 9.3 (SAS Institute, Cary, NC, USA).
Of the 195 patients, 32 (16%) had positive influenza test results. Patients who had a positive influenza test were more likely to receive quinolones (odds ratio [OR] 3.30, 95% CI 1.51-7.21) than were patients whose influenza tests were negative (Table 1). Patients who had a positive influenza test were significantly more likely to have the positive S. aureus respiratory culture collected <2 days after hospital admission than were the patients whose influenza tests were negative (OR 3.27, 95% CI 1.39-7.70).
Of the 32 influenza-positive patients, 9 (28%) died; of the 163 influenza-negative patients, 18 (11%) died (OR 3.15, 95% CI 1.27-7.86; p = 0.021) ( Table 2). Of the 9 influenza-positive patients who died, 5 had MRSA. Among the 27 patients who died, those with a positive influenza test were more likely to have diabetes than those who had a negative influenza test (33% vs. 0%; p = 0.029). The multivariable logistic regression model found that, after statistically adjusting for year and time from admission to collection of S. aureus culture samples, patients whose influenza tests were positive had >4-fold increased odds of death compared with patients whose influenza tests were negative (OR 4.31, 95% CI 1.57-11.83; p<0.005) ( Table 2).

Conclusions
Our results are consistent with the results of other studies. Other investigators reported poor outcomes among patients who were co-infected with influenza viruses and S. aureus (3,4,7). Kallen et al. found a statistically significant increased risk for death among patients who had positive influenza test results and community-acquired S. aureus pneumonia, compared with patients who had negative influenza test results and community-acquired S. aureus pneumonia (7). The Kallen et al. study included patients who had either MRSA or methicillin-susceptible S. aureus pneumonia (7) but evaluated only 47 patients. The sample size for our study was much larger than previously performed studies, and we were able to examine mortality rates among patients who had a respiratory culture that grew either MRSA or methicillin-susceptible S. aureus. 1254 Emerging Additionally, co-infection with influenza and S. aureus has been examined in animal models to identify mechanisms that cause poor outcomes (8)(9)(10)(11)(12). Severity of illness related to co-infection has been associated with a dysfunctional cell repair system and an altered immunologic response such as suppression of macrophage function, inhibition in phagocytic bacterial clearance, and cell damage to the airway system (8)(9)(10)(11)(12). Investigators have hypothesized that influenza damages epithelial cells in the respiratory system, providing opportunity for enhanced bacterial attachment (8,11). Once bacteria invade, cell destruction and fluid cause dysfunction of the airway system (8,11).
This study had limitations. First, the investigation might have excluded patients who were tested for influenza at other facilities or who did not have laboratoryconfirmed influenza. Second, we could not determine whether the respiratory cultures that grew S. aureus represented infections or colonization. However, the information we describe remains clinically relevant because often clinicians do not know whether patients with positive S. aureus cultures are infected or colonized. Diagnosing S. aureus pneumonia is challenging, and acquiring a lower respiratory culture such as a bronchial specimen or tracheal aspirate can be invasive and difficult to collect. Therefore, if S. aureus pneumonia is suspected (e.g., symptoms and positive sputum culture), patients may be treated without a confirmed positive lower respiratory culture. Third, our dataset did not include information about variables such as influenza vaccination status, mechanical ventilation, co-infection with organisms other than influenza and S. aureus, and whether the pneumonia was necrotizing. Fourth, misclassification bias may exist based on our definition of influenza infection. Patients with a negative influenza test may be misclassified since we were unable to determine the time interval between the onset of ILI symptoms and the collection of the influenza sample. Therefore, patients may have recovered from influenza before receiving an influenza test. Last, influenza-like illness ICD-9-CM codes were used to identify the cohort because the patients initially were included in a study of influenza-like illness and S. aureus pneumonia (J.S. Mc-Danel, unpub. data). Therefore, patients may have been missed if they had a respiratory infection with S. aureus and the condition or symptoms were not captured through an ICD-9-CM code.
In conclusion, among patients whose respiratory cultures grew S. aureus, patients with influenza were significantly more likely to die than were patients whose influenza tests were negative. Interventions that increase influenza vaccination rates among patients at high risk for S. aureus respiratory infections may prevent both coinfection and death.